From 1 January to 5 October 2020 there were 4055 confirmed cases of COVID-19 diagnosed in NSW; their mean age was 44.7 years (SD 20.2), 49.8% (n = 2019) were female, 84.2% (n = 3413) were resident in a major city and 34.7% (n = 1406) were classified in the lower 50% of socioeconomic deciles (7.2%, n = 290, were missing socioeconomic classification).
Of the 4055 confirmed cases, one was excluded from analyses (due to missing hospitalisation date). Of the remaining 4054 cases, 447 were admitted to hospital as inpatients following COVID-19 onset, of whom 34 died from COVID-19. Another 21, mostly from residential aged care facilities, were recorded as having died from COVID-19 but were not admitted to hospital, so in total 468 were classified as having a severe COVID-19 outcome (hospitalisation and/or death). Of the 468, 190 were either admitted to ICU or died and classified as having very severe COVID-19 outcomes.
As shown in Fig. 1 for a) severe COVID-19 and b) very severe COVID-19 there is a clear increase in the incidence of poor COVID-19 outcomes with age, with adults 70+ years much more likely to be hospitalised, admitted to ICU or to die from COVID-19.
Figure 2 shows the proportion of COVID-19 cases classified as severe by age, sex, socioeconomic status and area of residence, as well as the adjusted hazard ratios. Among adults, increasing age was the characteristic most strongly associated with severe COVID-19. After adjustments, compared to those aged 30–39 years the risk of severe COVID-19 was almost 10 times greater in those aged 80+ years and in those aged 60–79 years it was about 3 times higher. There was no significant effect of sex, with men and women having similar risk of COVID-19 hospitalisation or death (aHR 1.11, 95%CI 0.92–1.33). There was also a strong effect, only in the most socioeconomically advantaged group of reduced risk of severe COVID-19 (aHR 0.54, 95%CI 0.37–0.81).
Figure 3 shows the numbers and proportions of cases with severe COVID-19 according to comorbidities, and the adjusted hazard ratios comparing those with the comorbidity to those without. Of those without a comorbidity, around 5% were hospitalised or died. This compares to 49% and 53% in those with cerebrovascular disease or chronic kidney disease. However after adjusting for age, sex, socioeconomic status and the presence of other comorbidities, these risks were attenuated. In fully adjusted models, the highest hazard ratios for a single comorbidity were in people with diabetes, aHR 1.93 (95%CI 1.52–2.45) compared to those without diabetes, with COPD/bronchitis compared to those without, aHR 1.81 (95%CI 1.43–2.29), and for those with immunosuppressive conditions aHR 1.66 (95%CI 1.19–2.33) compared to those without. Having any of the examined comorbidities compared to none was also associated with increased risk with aHRs of 2.33 (95%CI 1.81–2.99). Neither being a current smoker or pregnancy was found to be associated with increased risk of hospitalisation or death from COVID-19 although the numbers of pregnant women diagnosed with COVID-19 was relatively small.
Figures 4 and 5 show hazard ratios for very severe COVID-19 (ICU admission or death) based on 190 events among 4054 cases. Results were somewhat similar to those shown in Figs. 2 and 3 examining hospitalisation or death. Older age remained by far the predominant risk factor. Hazard ratios comparing those aged 20–29 to those aged 30–39 years were 0.27 (95%CI 0.10–0.74). Hazard ratios then increased exponentially to 4.45 (95%CI 2.49–7.97) in those 70–79, 8.43 (95%CI 4.44–16.03) in those aged 80–89 years and 16.19 (95%CI 7.77–33.76) in those 90+ years. Men had higher risks for very severe COVID-19 than women (aHR 1.40, 95%CI 1.04–1.88). Also, those who were the least disadvantaged had only half the risk of very severe COVID-19 as those who were most disadvantaged (aHR 0.48, 95%CI 0.29–0.80). For ICU or death from COVID-19, diabetes, immunosuppressive conditions, obesity, and COPD/chronic bronchitis, had the strongest associations. A recent cancer diagnosis and chronic kidney disease were also found to have elevated point estimates for risk of the same magnitude as the other conditions, but numbers were relatively small and fully adjusted hazard ratios were not statistically significant. There were insufficient numbers of smokers and pregnant women who were admitted to ICU or who died from COVID-19 to calculate risk ratios.
Figure 6 shows that risks of both severe and very severe COVID-19 increased significantly with the number of comorbidities. Compared to those without any comorbidities, risks increased linearly with increasing numbers of comorbidities to about 4–5 times higher in those with 3 or more comorbidities.
In sensitivity analyses in which the population was restricted to cases who acquired their infections locally (N = 1795) there were no significant deviations from the main findings although for some groups the numbers of cases and events were substantially smaller (data not shown). When we restricted analyses to cases who had an onset date from 1 March 2020, the majority of cases were included in analyses (4004/4054, 99%) so findings were unchanged.